Fundamentals of Voice Recognition | Between2Beats

Fundamentals of Voice Recognition

Voice recognition is something that has been around for a couple of decades now and yet people still wonder about some aspects of it. If you are wondering about certain aspects of voice recognition, such as its origin or how it can simplify how we use our technology, this article will answer those questions, among other aspects.

Based on S01E13 of the Between2Beats Live to Stream Podcast Series

The Origins of Voice Recognition

Voice commands are a means of accessing a back-end process. The back-end process can be on the internet, in the Cloud, on your smartwatch, or on your phone. The process really is the money maker. The more we incorporate that process into an automated command, and the more we use automated commands, the more that will become how we do it.?

You have means now to issue auditory commands with such devices as a Google Home or an Amazon Echo. But that’s just an auditory command. We are at a point now where back-end systems have done enough processing and analysis to get a better understanding of what we’re trying to do.?

How has the technology changed over the years from simple commands to voice texting?

With voice, what’s really changed in the last 20 years, is that the systems have opened up. All of the local usage goes to a large database that allows us to make sense of it and make it easier.

It’s interpreting what you’re saying as a mix of a million people saying the same thing, and saying that you’re in the range. You’re in that sweet spot of the command, for example, that includes, “text Mom.” However, it could also include, “text Momma,” or “check Mom.”

The A.I. and big data come in behind the scenes with all of these examples over time. It has categorized it by saying this grouping of variations of this command all go “here.” When you start stripping that tech down, your phone or your smartwatch has just made it easier for you to access that backend Cloud interface with millions of interpretations.

In the early days, all of that processing was done on the phone. It wasn’t connected to a backend Cloud. All it could figure out was all you could cram intelligence-wise on a little chip on that device. Now that processing and database that makes up all of those variations exists somewhere else.

If you’re logged into your device, now all of a sudden your profile knows when you’re asleep, it knows when you’re awake, etc. It incorporates that into the vocal command, so it now has context. If you have a set schedule, the assistant layer will go through your profile and your data usage and it figures out patterns. When you ask for something, it may know that you may want something different in the morning or at night.?

Is sending all of our commands to the Cloud better, having the option of big data, than local usage directly from our devices?

We need it. Over the last five to 10 years has drastically improved voice recognition. The improvements in A.I., avatars and even Siri, have only been possible because of big data. We could not achieve big data, if we did not send all of our data to multiple systems.?

The difference between Voice Command and Voice Search

Everything is a command, or series of commands, whether it’s a voice command or voice search. The big difference is searching is more conversational and uses more natural language than a command. Voice search is less about the technology and more about how everything comes together to say “what is your natural way of asking for this information?” It’s not typing it. It’s asking for it. At the end of the day, voice search is issuing a voice command using natural language as well as a larger pool of data to curate the result due to the many more variations of the words spoken into the device.?